The Broadly Selective Human Na (cid:1) /Nucleoside Cotransporter (hCNT3) Exhibits Novel Cation-coupled Nucleoside Transport Characteristics*

The concentrative nucleoside transporter (CNT) protein family in humans is represented by three members, hCNT1, hCNT2, and hCNT3. hCNT3, a Na (cid:1) /nucleoside symporter, transports a broad range of physiological purine and pyrimidine nucleosides as well as anticancer and antiviral nucleoside drugs, and belongs to a different CNT subfamily than hCNT1/2. H (cid:1) -dependent Escherichia coli NupC and Candida albicans CaCNT are also CNT family members. The present study utilized heterologous expression in Xenopus oocytes to investigate the specificity, mechanism, energetics, and structural basis of hCNT3 cation coupling. hCNT3 exhibited uniquely broad cation interactions with Na (cid:1) , H (cid:1) , and Li (cid:1) not shared by Na (cid:1) coupled hCNT1/2 or H (cid:1) -coupled NupC/CaCNT. Na (cid:1) medium m M Studies— Nucleoside-evoked membrane currents were measured in hCNT3-producing oocytes at room temperature (20 °C) using a GeneClamp 500B oocyte clamp (Axon Instruments Inc., Foster City, CA) in the two-electrode, voltage clamp mode. The Gene- Clamp 500B was interfaced to an IBM-compatible PC via a Digidata 1200A/D converter and controlled by pCLAMP software (version 8.0, Axon Instruments Inc.). The microelectrodes were filled with 3 M KCl and had resistances that ranged from 0.5 to 2.5 megohms. Oocytes were penetrated with the microelectrodes, and their membrane potentials were monitored for periods of 10–15 min. Oocytes were discarded when membrane potentials were unstable, or more positive than (cid:5) 30 mV. Unless otherwise indicated, the oocyte membrane potential was clamped at a holding potential ( V h ) of (cid:5) 50 mV, and nucleoside was added in the appropriate transport medium at a concentration of 100 (cid:3) M . Current signals were filtered at 20 Hz (four-pole Bessel filter) and sampled at a sampling interval of 20 ms. For data presentation, the signals were further filtered at 0.5 Hz by the pCLAMP program suite. more closely related in sequence to the prevertebrate hagfish transporter hfCNT (30) than to mammalian CNT1/2, and thus form a separate CNT subfamily. The present study utilized heterologous expression in Xenopus oocytes in combination with electrophysiological, radioisotope flux, and chimeric experiments to characterize the selectivity, mechanism, energet-ics, and structural basis of hCNT3 cation coupling. Parallel experiments with mCNT3 confirmed the general applicability of the reported findings.

Physiological nucleosides and synthetic nucleoside analogs have important biochemical, physiological, and pharmacologi-cal activities in humans. Adenosine, for example, has purinoreceptor-mediated functions in processes such as modulation of immune responses, platelet aggregation, renal function, and coronary vasodilation (1,2). Nucleosides also provide salvage precursors for nucleic acid biosynthesis, and nucleoside drugs are commonly used in the therapy of cancer and viral infections (3,4). Most nucleosides, including those with antineoplastic and/or antiviral activities, are hydrophilic and require specialized plasma membrane nucleoside transporter (NT) 1 proteins for their uptake into or release from cells (5)(6)(7).
Multiple transport systems for nucleosides have been observed in human and other mammalian cells and tissues (7)(8)(9). The concentrative systems (cit, cif, and cib) 2 are inwardly directed Na ϩ -dependent processes present primarily in intestinal and renal epithelia and other specialized cells (7)(8)(9). The equilibrative systems (es and ei) mediate bidirectional downhill transport of nucleosides, have generally lower substrate affinities than the concentrative systems, and occur in most, possibly all, cell types (7)(8)(9). Systems cit and cif transport adenosine and uridine, but are otherwise pyrimidine and purine nucleoside-selective, respectively, whereas systems cib, es, and ei transport both pyrimidine and purine nucleosides. System es is inhibited by nanomolar concentrations of nitrobenzylthioinosine (NBMPR), whereas system ei also transports nucleobases (7)(8)(9)(10).
Human and mouse CNT3 (hCNT3 and mCNT3) are the most recent mammalian CNT family members to be identified (18) and, together with cib-type hagfish CNT (hfCNT) (30), form a phylogenetic CNT subfamily separate from the mammalian CNT1/2 subfamily (18). In addition to differences in substrate selectivities, members of the two subfamilies also differ in the stoichiometry of Na ϩ /nucleoside coupling. Contrary to a recent report in this journal (31), for example, we have shown that hCNT1 has a Na ϩ /nucleoside coupling ratio of 1:1 (32), whereas the coupling ratio of hfCNT is 2:1 (30). Here, we extend these investigations and present a mechanistic and chimeric analysis of cation coupling in hCNT3. The results validate our previously reported differences in cation coupling between the CNT3/hfCNT and CNT1/2 subfamilies and reveal additional novel features of CNT3-cation interactions.

EXPERIMENTAL PROCEDURES
hCNT3 cDNA-cDNA encoding hCNT3 (GenBank TM accession number AF305210) in the enhanced Xenopus plasmid expression vector pGEM-HE (33) with flanking 5Ј-and 3Ј-untranslated regions from the Xenopus ␤-globin gene was obtained as previously described (18).
In Vitro Transcription and Expression in Xenopus Oocytes-hCNT3 plasmid DNA was linearized with NheI and transcribed with T7 polymerase using the mMESSAGE mMACHINE TM (Ambion, Austin, TX) transcription system. The remaining template was removed by digestion with RNase-free DNase1. Stages V-VI oocytes from Xenopus laevis were treated with collagenase (2 mg/ml) for 2 h, and remaining follicular layers were removed by phosphate treatment (100 mM K 2 PO 4 ) and manual defolliculation (11). Twenty-four hours after defolliculation, oocytes were injected with either 10 nl of water containing 10 ng of RNA transcript encoding hCNT3 or 10 nl of water alone. Injected oocytes were then incubated for either 4 days (radioisotope flux studies) or 4 -7 days (electrophysiology) at 18°C in modified Barth's solution (changed daily) (88 mM NaCl, 1 mM KCl, 0.33 mM Ca(NO 3 ) 2 , 0.41 mM CaCl 2 , 0.82 mM MgSO 4 , 2.4 mM NaHCO 3 , 10 mM Hepes, 2.5 mM sodium pyruvate, 0.05 mg/ml penicillin, and 0.1 mg/ml gentamycin sulfate, pH 7.5).
Transport Media-Unless otherwise stated, electrophysiological and radioisotope flux studies used Na ϩ -containing transport medium composed of 100 mM NaCl, 2 mM KCl, 1 mM CaCl 2 , 1 mM MgCl 2 , and 10 mM Hepes (for pH values Ͼ 6.5) or 10 mM MES (for pH values Յ 6.5). In experiments examining the Na ϩ dependence of transport (i.e. where the indicated Na ϩ concentration is Ͻ100 mM), Na ϩ in the transport medium was replaced by equimolar choline chloride (ChCl) to maintain isomolarity. Proton dependence was tested in Na ϩ -free choline-containing transport medium (100 mM ChCl) at pH values ranging from 4.5 to 8.5. Experiments testing Li ϩ coupling of transport were performed in medium at pH 8.5 containing Li ϩ (100 mM LiCl) in place of Na ϩ .
Electrophysiological Studies-Nucleoside-evoked membrane currents were measured in hCNT3-producing oocytes at room temperature (20°C) using a GeneClamp 500B oocyte clamp (Axon Instruments Inc., Foster City, CA) in the two-electrode, voltage clamp mode. The Gene-Clamp 500B was interfaced to an IBM-compatible PC via a Digidata 1200A/D converter and controlled by pCLAMP software (version 8.0, Axon Instruments Inc.). The microelectrodes were filled with 3 M KCl and had resistances that ranged from 0.5 to 2.5 megohms. Oocytes were penetrated with the microelectrodes, and their membrane potentials were monitored for periods of 10 -15 min. Oocytes were discarded when membrane potentials were unstable, or more positive than Ϫ30 mV. Unless otherwise indicated, the oocyte membrane potential was clamped at a holding potential (V h ) of Ϫ50 mV, and nucleoside was added in the appropriate transport medium at a concentration of 100 M. Current signals were filtered at 20 Hz (four-pole Bessel filter) and sampled at a sampling interval of 20 ms. For data presentation, the signals were further filtered at 0.5 Hz by the pCLAMP program suite.
In protonophore studies, carbonyl cyanide m-chlorophenylhydrazone (CCCP) (100 M) was preincubated with oocytes for 15 min prior to measuring uridine-evoked currents (100 M) in Na ϩ -free medium (100 mM ChCl, pH 5.5). Stock solutions of CCCP were dissolved in Me 2 SO. Control experiments confirmed that oocytes were unaffected by Me 2 SO at its final concentration of 0.5% (w/v).
Current-voltage (I-V) curves were determined from differences in steady-state currents generated in the presence and absence of substrate during 300-ms voltage pulses to potentials between ϩ60 and Ϫ110 mV (10-mV steps). For I-V relations, currents were filtered at 2 kHz (four-pole Bessel filter) and sampled at a rate of 200 s/point (corresponding to a sampling frequency of 5 kHz).
Data from individual electrophysiology experiments are presented as nucleoside-evoked currents from single representative cells or as mean values (ϮS.E.) from four or more oocytes from the same batch of oocytes used on the same day. Each experiment was repeated at least twice on oocytes from different frogs. No nucleoside-evoked currents were detected in oocytes injected with water alone, demonstrating that currents in hCNT3-producing oocytes were transporter-specific.
Radioisotope Flux Studies-Radioisotope transport assays were performed as described previously (11,18) on groups of 10 -12 oocytes at 20°C using 14 C-labeled nucleosides (1 Ci/ml) in 200 l of the appropriate transport medium. Unless otherwise stated, nucleoside uptake was determined at a concentration of 20 M. Following incubation, seven rapid washes in ice-cold choline chloride transport medium (pH 7.5) removed extracellular label, and individual oocytes were dissolved in 1% (w/v) SDS for quantitation of cell-associated radioactivity by liquid scintillation counting (LS 6000 IC, Beckman, Fullerton, CA). Uptake values represent initial rates of transport (1-min flux) and are presented as means Ϯ S.E. of 10 -12 oocytes from representative experiments. Individual experiments were performed on cells from single batches of oocytes used on the same day. Transporter-mediated fluxes were calculated as uptake in RNA transcript-injected oocytes minus uptake in control water-injected oocytes. Each experiment was repeated at least twice using oocytes from different frogs.
Kinetic Parameters-Kinetic parameters calculated from electrophysiological and radioisotope flux experiments were determined by least squares fits to the Michaelis-Menten and Hill equations using Sigmaplot 2000 software (Jandel Scientific Software, San Rafael, CA). Those from electrophysiological experiments were determined from fits to data from individual oocytes normalized to the I max value obtained for that oocyte, and are presented as values (ϮS.E.) for single representative oocytes, or as means (ϮS.E.) of four or more cells. Those from radioisotope experiments were derived from curve fits to averaged mediated data from 10 -12 oocytes, and are presented as means (ϮS.E.).
Charge-to-Nucleoside Stoichiometry-Na ϩ /nucleoside and H ϩ /nucleoside coupling ratios for hCNT3 were determined by simultaneously measuring Na ϩ or H ϩ currents and [ 14 C]uridine (200 M, 1 Ci/ml) uptake under voltage clamp conditions. Individual hCNT3-producing oocytes were placed in a perfusion chamber and voltage clamped at V h of Ϫ30, Ϫ50, or Ϫ90 mV in the appropriate nucleoside-free medium for a 10-min period to monitor baseline currents. When the baseline was stable, the nucleoside-free medium was exchanged with medium of the same composition containing radiolabeled uridine. Current was measured for 3 min, and uptake of uridine was terminated by washing the oocyte with nucleoside-free medium until the current returned to baseline. The oocyte was then transferred to a scintillation vial and solubilized with 1% (w/v) SDS for quantitation of oocyte-associated radioactivity. Nucleoside-induced current was obtained as the difference between baseline current and the inward uridine current. The total charge translocated into the oocyte during the uptake period was calculated from the current-time integral and correlated with the measured radiolabeled flux for each oocyte to determine the charge/uptake ratio. Basal [ 14 C]uridine uptake was determined in control water-injected oocytes (from the same donor frog) under equivalent conditions and used to correct for endogenous non-mediated uridine uptake over the same incubation period. Coupling ratios (ϮS.E.) were calculated from slopes of least-squares fits of uridine-dependent charge versus uridine accumulation for eight or more oocytes.
Charge-to-Na ϩ Stoichiometry-hCNT3-mediated uptake of 22 Na ϩ was optimized for specific activity and transport rate by using a saturating concentration of uridine (200 M) and a 22 Na ϩ concentration of 1 mM, a value close to the K 50 for Na ϩ . Individual hCNT3-producing oocytes were voltage clamped at Ϫ90 mV and perfused with Na ϩ -free medium (100 mM ChCl, pH 8.5). A stable baseline current was recorded, and the bath solution was changed for 3 min to one containing 200 M uridine, 1 mM 22 Na ϩ (1 Ci/ml), and 99 mM ChCl (pH 8.5). The solution was changed back to uridine-and Na ϩ -free medium until the current returned to baseline. The oocyte was then transferred to a scintillation vial and solubilized with 1% (w/v) SDS for quantitation of oocyteassociated radioactivity. Testing of each individual hCNT3-producing oocyte prior to addition of uridine and 22 Na ϩ showed no shift in the baseline current when the composition of the bath solution was changed from 100 mM ChCl to 1 mM NaCl plus 99 mM ChCl, indicating an absence of detectable hCNT3 Na ϩ slippage under the experimental conditions used. Similarly, basal 22 Na ϩ uptake in control water-injected oocytes (from the same donor frog) was determined under equivalent conditions over the same incubation period and used to correct for endogenous non-mediated Na ϩ uptake. For each hCNT3-producing oocyte, the total charge translocated during the uptake period was calculated from the current-time integral and correlated with the measured 22 Na ϩ flux to determine charge/uptake ratio. The charge-to-Na ϩ stoichiometry (ϮS.E.) was calculated from the slope of a least-squares fit of uridine-dependent charge versus 22 Na ϩ accumulation for five oocytes.
Construction of Chimeric hCNT3 and hCNT1 Transporters-hCNT1 cDNA (GenBank TM accession number U62968) (14) was subcloned into the vector pGEM-HE prior to chimera construction with hCNT3 (already in pGEM-HE) to enhance expression in Xenopus oocytes. Two sets of overlap primers were designed at a splice site between Lys 314 and Val 315 of hCNT3 and the corresponding residues in hCNT1 (Lys 293 and Ile 294 ) in the putative loop linking TM 6 and TM 7 (arrow in Fig. 11A) to create reciprocal 50:50 chimeras by a two-step overlap extension PCR method (30,34). Chimeric constructs containing the restriction site KpnI downstream of the M13 forward primer and the restriction site SphI upstream of the M13 reverse primer were subcloned into the respective restriction sites of the pGEM-HE vector. The chimeras were sequenced in both directions to verify the splice sites and ensure that no mutations had been introduced.
Chemicals-Nucleosides and CCCP were purchased from Sigma. The 14 C-labeled nucleosides were purchased from Moravek Biochemicals (Brea, CA) or Amersham Biosciences. 22 Na ϩ was obtained from Amersham Biosciences.

RESULTS
Cation Dependence of hCNT3-We have previously used heterologous expression in Xenopus oocytes in combination with radioisotope flux assays and the two-microelectrode voltage clamp to demonstrate that recombinant hCNT3 functions as a broad specificity cib-type electrogenic Na ϩ /nucleoside symporter (18). The experiment of Fig A, representative cation/nucleoside current traces in a single hCNT3-producing oocyte clamped at Ϫ50 mV in transport medium containing either Na ϩ (100 mM NaCl, pH 8.5), choline (100 mM ChCl, pH 5.5), or Li ϩ (100 mM LiCl, pH 8.5). The bar denotes addition of uridine (100 M) to the bath. No currents were observed in control water-injected oocytes. B, averaged inward currents in hCNT3-producing oocytes perfused sequentially with 100 M uridine in Na ϩ -containing transport medium (black bar; 100 mM NaCl, pH 8.5), Na ϩ -free transport medium (open bar; 100 mM ChCl, pH 5.5), and Na ϩ -free transport medium containing 100 M CCCP (gray bar; 100 mM ChCl, pH 5.5). Currents are means Ϯ S.E. of five different oocytes from the same batch of cells used on the same day.
initial rate of hCNT3-mediated uptake (influx) of [ 14 C]uridine (20 M) measured at extracellular pH values ranging from 5.5 to 8.5 under Na ϩ -free conditions. Choline (Ch ϩ ) was substituted for Na ϩ , and values were corrected for basal non-mediated uptake in control water-injected oocytes (Ͻ0.03 pmol/ oocyte⅐min Ϫ1 under all conditions tested). A marked pH dependence of uridine influx was apparent. In 100 mM ChCl at pH 5.5, for example, hCNT3-mediated uridine influx was 26fold higher than at pH 8.5, and was ϳ60% of that obtained in the presence of 100 mM NaCl at pH 7.5. In the absence of either H ϩ or Na ϩ , hCNT3 also exhibited Li ϩ -dependent uridine influx (100 mM LiCl, pH 8.5, versus 100 mM ChCl, pH 8.5). In marked contrast, hCNT1 and hCNT2, the two other human CNT isoforms, showed no pH-dependent uridine uptake and exhibited very small Li ϩ -mediated uridine influx (Յ2% of the corresponding Na ϩ -mediated uridine flux; data not shown). As illustrated in Fig. 2, the novel H ϩ and Li ϩ dependence of hCNT3 were further investigated by electrophysiology. External application of 100 M uridine to a representative hCNT3-producing oocyte clamped at Ϫ50 mV under Na ϩ -free conditions elicited inward H ϩ and Li ϩ currents that returned to baseline upon removal of substrate ( Fig. 2A). As demonstrated by the mean current data in Fig. 2B, H ϩ currents were markedly inhibited by pre-treatment of oocytes with the protonophore CCCP. No currents were FIG. 3. Substrate selectivity and pH dependence of nucleoside transport by recombinant hCNT3. Oocytes were injected with either water alone or water containing hCNT3 RNA transcripts. A, averaged nucleoside-induced inward currents measured by perfusing hCNT3producing oocytes with 100 M pyrimidine (uridine, thymidine, and cytidine) or purine (adenosine, inosine, and guanosine) nucleosides in Na ϩ -containing transport medium (100 mM NaCl, pH 7.5). Currents are means Ϯ S.E. of 5-6 different oocytes from the same batch of cells used on the same day. B, mean nucleoside-induced inward currents measured by perfusing hCNT3-producing oocytes with 100 M pyrimidine and purine nucleosides in Na ϩ -free transport medium (100 mM ChCl, pH 5.5 and 8.5). Currents are means Ϯ S.E. of 5-6 oocytes from the same batch of cells used on the same day. C, mediated radiolabeled fluxes of pyrimidine and purine nucleosides (20 M, 20°C, 1-min flux) in oocytes injected with hCNT3 RNA transcripts measured in Na ϩ -free transport media (100 mM ChCl, pH 5.5 and 8.5). Mediated transport was calculated as uptake in RNA-injected oocytes minus uptake in control water-injected oocytes. Each value represents the mean Ϯ S.E. of 10 -12 oocytes obtained from the same batch of cells and used on the same day. D, mean nucleoside-induced inward currents measured by perfusing hCNT3-producing oocytes with cytidine, inosine, or guanosine (100 M and 1 mM) in Na ϩ -free transport medium (100 mM ChCl, pH 5.5 and 8.5). Currents are means Ϯ S.E. of 5-6 oocytes from the same batch of cells used on the same day. In A, B, and D, no currents were observed in control water-injected oocytes. observed in control water-injected oocytes (data not shown). Thus, in addition to being Na ϩ -dependent, hCNT3 functioned as an electrogenic H ϩ /nucleoside and Li ϩ /nucleoside symporter. These findings were also confirmed in parallel studies of mCNT3 (data not shown). Subsequent experiments focused on the mechanism(s) of Na ϩ and H ϩ coupling of hCNT3.
Cation-induced Changes in hCNT3 Substrate Selectivity-In Na ϩ -containing medium at physiological pH 7.5, hCNT3 transports different radiolabeled physiological pyrimidine and purine nucleosides with similar apparent K m values and V max :K m ratios (18). As illustrated by the mean current data in Fig. 3A, a panel of pyrimidine (uridine, thymidine, and cytidine) and purine (adenosine, inosine, and guanosine) nucleosides elicited similar large inward Na ϩ currents when applied at 100 M to hCNT3-producing oocytes in NaCl transport medium at pH 7.5. The corresponding inward currents generated by the same panel of nucleosides under Na ϩ -free conditions were measured in ChCl transport medium of increasing acidity (pH 8.5, 7.5, 6.5, and 5.5). For clarity of presentation, only values obtained at pH 5.5 and 8.5 are shown in Fig. 3B. Similar to the trend seen for uridine in Fig. 1, inward pH-dependent nucleosideevoked currents were evident for uridine, thymidine, and adenosine, were less marked for cytidine and inosine, and were absent for guanosine. A corresponding selectivity profile for hCNT3 H ϩ /nucleoside cotransport was obtained in radioisotope nucleoside influx assays (Fig. 3C). Those nucleosides in Fig. 3 (B and C) that exhibited the lowest transport activity in Na ϩfree acidified medium (cytidine, inosine, and guanosine) were also tested by electrophysiology at a higher substrate concentration of 1 mM (Fig. 3D). Substrate-induced H ϩ currents were confirmed for cytidine and inosine, whereas guanosine-evoked currents remained low and pH-independent. For all nucleosides and conditions tested, no currents were evident (data not shown) and basal non-mediated radioisotope fluxes were Ͻ0.07 pmol/oocyte⅐min Ϫ1 in control water-injected oocytes. Na ϩ -and H ϩ -coupled hCNT3 therefore exhibited markedly different se-  (1 mM) in an hCNT3-producing oocyte in transport medium containing either Na ϩ (100 mM NaCl, pH 7.5) or choline (100 mM ChCl, pH 5.5). Uridine-evoked currents in the same hCNT3-producing oocyte were 210 and 90 nA in Na ϩ -containing and cholinecontaining transport media, respectively. B, a comparison of hCNT3-mediated inward currents following sequential addition of uridine (100 M), gemcitabine (100 M), AZT (1 mM), or ddC (1 mM) to transport media containing either Na ϩ (100 mM NaCl, pH 7.5) or choline (100 mM ChCl, pH 5.5). Currents are means Ϯ S.E. for three different oocytes from the same batch of cells used on the same day. No currents were observed in control waterinjected oocytes. lectivity profiles for physiological nucleosides.
This difference in substrate selectivity between Na ϩ -and H ϩ -coupled hCNT3 extended to interactions with therapeutic nucleoside analogs. In Na ϩ -containing medium at physiological pH 7.5, hCNT3 efficiently transports the anticancer nucleoside drug gemcitabine, and mediates lower, but still significant, fluxes of the antiviral nucleoside drugs AZT, ddC, and ddI (18). As shown by the traces from a single representative oocyte in Fig. 4A and by the mean current data in Fig. 4B, gemcitabine (100 M) elicited inward Na ϩ and H ϩ currents that returned to baseline upon removal of substrate, whereas ddC and AZT (1 mM) produced inward currents only in the presence of Na ϩ . No currents were detected in control water-injected oocytes (data not shown). The inability of H ϩ -coupled hCNT3 to support ddC and AZT transport was confirmed in parallel radioisotope flux experiments (data not shown).
Voltage Dependence of hCNT3 Transport Currents- Fig. 5A shows representative current traces in an hCNT3-producing oocyte in Na ϩ -containing medium (pH 8.5) before and after perfusion with 100 M uridine in an experiment undertaken to examine the effects of membrane potential on uridine-induced steady-state currents. Currents evoked by uridine at potentials between ϩ60 and Ϫ110 mV were voltage-dependent and increased as the membrane potential became more negative (Fig.  5B). Uridine-induced Na ϩ currents approached zero but did not reverse polarity at potentials up to ϩ60 mV. Measured in the same oocyte, uridine-induced H ϩ currents (ChCl, pH 5.5) were approximately half those in Na ϩ -containing medium and exhibited a similar voltage dependence. In other experiments performed over a wider V h range, uridine-induced Na ϩ and H ϩ currents did not saturate at negative potentials up to Ϫ150 mV (data not shown).
Voltage Dependence of hCNT3 Transport Kinetics-The effects of membrane potential on hCNT3 transport kinetics in the presence of external Na ϩ were examined in detail by measuring the apparent affinities for uridine (K m ) and Na ϩ (K 50 ) and the uridine-evoked maximum current (I max ) at four different holding potentials (V h ϭ Ϫ10, Ϫ30, Ϫ50, and Ϫ70 mV) (Fig.  6). K m was determined in transport medium containing either 10 or 100 mM NaCl (pH 8.5), and mean values plotted as a function of V h are summarized in Fig. 6C. As representative examples of the kinetic data used to generate K m values, Fig. 6, A and B, show uridine concentration dependence curves recorded from individual oocytes at a membrane potential of Ϫ10 mV and external Na ϩ concentrations of 10 and 100 mM, respectively. K 50 was determined at an external uridine concentration of 100 M (pH 8.5), and mean values plotted as a function of V h summarized in Fig. 6E. The representative single oocyte Na ϩ -activation curve shown in Fig. 6D was measured at a holding potential of Ϫ10 mV. I max values were determined independently at a saturating external uridine concentration of 100 M and a Na ϩ concentration of 100 mM (pH 8.5) (Fig. 6F).
As demonstrated in Fig. 6C, K m was unaffected by membrane potential at 100 mM external Na ϩ , but was voltage-dependent at 10 mM Na ϩ , decreasing from 24 to 7.3 M as the membrane potential was made more negative. At high negative potentials, the K m value at 10 mM NaCl approached that observed at 100 mM external Na ϩ , suggesting that the effect of membrane potential on K m was predominantly the result of voltage dependence of Na ϩ binding (35,36). Consistent with this, we found that the Na ϩ K 50 value decreased from 5.3 mM at Ϫ10 mV to 2.1 mM at Ϫ70 mV (Fig. 6E). As shown in Fig. 6F, I max also increased as the membrane potential was made more negative. At a holding potential of Ϫ10 mV (100 mM NaCl), the FIG. 5. Voltage dependence of hCNT3-mediated currents. A, time course of transmembrane currents recorded from a representative hCNT3-producing oocyte in the presence of 100 M external uridine (100 mM NaCl, pH 8.5) in response to 300-ms voltage pulses from a holding potential of Ϫ50 mV (left trace). Currents are shown at four different test potentials only (V t ϭ ϩ50, Ϫ10, Ϫ60, and Ϫ110 mV). The capacitive transients have been truncated to clearly demonstrate the steady-state currents. The right trace shows corresponding currents from the same oocyte recorded in the same transport medium in the absence of uridine. The records are offset to zero. B, the current-voltage (I-V) curves for hCNT3 were generated from the difference between steady-state currents recorded in the presence and absence of 100 M uridine in Na ϩ -containing (open circles; 100 mM NaCl, pH 8.5) or choline-containing (solid circles; 100 mM ChCl, pH 5.5) transport media upon voltage pulses from V h of Ϫ50 mV to final potentials ranging between ϩ60 and Ϫ110 mV, in 10-mV steps. The data are from the same oocyte as in A. No uridine-induced currents were observed in control water-injected oocytes. mean uridine-evoked inward current was 37 nA. This increased to 97 nA at Ϫ70 mV, a trend that was similar to the I-V relationship shown in Fig. 5B. Calculated over the same limited V h range (Ϫ10 to Ϫ70 mV), I-V curves for five oocytes gave an e-fold change (ϮS.E.) in current per 77 Ϯ 4 mV, compared with 67 Ϯ 11 mV for the data in Fig. 6F. I max reflects movement of the loaded and empty carrier (36). Similar to the Na ϩ /glucose cotransporter SGLT1, therefore, membrane potential influenced both ion binding and carrier translocation (35,36).
Cation Dependence of Uridine Transport Kinetics- Fig. 7 shows the concentration dependence of [ 14 C]uridine influx in Na ϩ -and choline-containing transport media at pH 5.5 and 7.5 FIG. 6. Voltage dependence of the transport kinetics of hCNT3. A, uridine concentration-response curve measured in a single representative hCNT3-producing oocyte at an external Na ϩ concentration of 10 mM (pH 8.5) and a membrane potential of Ϫ10 mV. Currents at each uridine concentration were normalized to the fitted I max value for that oocyte (ϮS.E.) of 52 Ϯ 1 nA. The K m value was 32 Ϯ 3 M. B, recordings from a single representative hCNT3-producing oocyte demonstrating a corresponding experiment at the same membrane potential (Ϫ10 mV) in the presence of 100 mM external Na ϩ (pH 8.5). Currents were normalized to the fitted I max of 55 Ϯ 3 nA. The K m was 7.6 Ϯ 1.1 M. C, the effect of membrane potential on uridine K m was determined at external Na ϩ concentrations of 10 mM (open circles) and 100 mM (solid circles) (pH 8.5) and holding potentials of Ϫ10, Ϫ30, Ϫ50, and Ϫ70 mV. K m values at each membrane potential were obtained from fits to data from individual oocytes normalized to the I max value obtained for that cell and are presented as means Ϯ S.E. of 4 -8 oocytes. D, Na ϩ concentration-response curve (pH 8.5) measured in a single representative hCNT3-producing oocyte at a membrane potential of Ϫ10 mV (100 M uridine). Currents at each Na ϩ concentration were normalized to the fitted I max value of 43 Ϯ 2 nA. The Na ϩ K 50 was 5.5 Ϯ 0.5 mM. E, the effect of membrane potential on Na ϩ K 50 was determined at holding potentials of Ϫ10, Ϫ30, Ϫ50, and Ϫ70 mV (100 M uridine, pH 8.5). K 50 values at each membrane potential were obtained from fits to data from individual oocytes normalized to the I max value obtained for that cell and are presented as means Ϯ S.E. of 5-7 oocytes. F, the effect of membrane potential on maximum current (I max ) was determined at holding potentials of Ϫ10, Ϫ30, Ϫ50, and Ϫ70 mV in the presence of 100 mM external Na ϩ (pH 8.5) and a saturating concentration of uridine (100 M). Each value is the mean Ϯ S.E. of 3-4 oocytes from the same batch of cells used on the same day. No currents were observed in control water-injected oocytes. Note: to more accurately determine K m and K 50 values (panels A-E), kinetic experiments at low membrane potentials were performed on preselected oocytes with maximal currents Ն 40 nA. The effect of membrane potential on I max (panel F) was measured independently in a separate experiment. measured in both hCNT3-producing and control water-injected oocytes. Kinetic parameters (K m and V max ) derived from these data for the hCNT3-mediated component of influx (uptake in RNA transcript-injected oocytes minus uptake in water-injected oocytes) are presented in Table I. Removal of extracellular Na ϩ at pH 7.5 led to a greater than 30-fold increase in the K m value for uridine influx from 17 to 580 M that was partially offset by a small (1.6-fold) increase in V max (Fig. 7, A and B). As shown in Fig. 7D, the decrease in uridine apparent affinity was substantially reversed by acidification of the transport medium to pH 5.5 (uridine K m 110 M). In contrast, acidification of the transport medium in the presence of Na ϩ had only modest effects on uridine transport kinetics (Fig. 7, A and C). V max :K m ratios, a measure of transporter efficiency, were as follows: 2.0 in the presence of Na ϩ (NaCl, pH 7.5), 0.09 in the absence of Na ϩ (ChCl, pH 7.5), 0.58 in the presence of H ϩ (ChCl, pH 5.5), and 1.7 with both cations (Na ϩ plus H ϩ ) present (NaCl, pH 5.5). Therefore, Na ϩ and H ϩ activated hCNT3 through mechanisms resulting in increased uridine apparent binding affinity. Relative to Na ϩ alone, Na ϩ plus H ϩ elicited no further shift in uridine K m , suggesting that the two cations exert their effects by binding to a common or overlapping site(s).
Na ϩ and H ϩ Activation Kinetics-The relationship between hCNT3-mediated uridine-evoked current and Na ϩ concentration (pH 8.5) was measured in oocytes clamped at Ϫ50 mV at three different uridine concentrations (5, 25, and 100 M). Kinetic parameters derived from these experiments are presented in Table II. As reported previously for hCNT3 (and mCNT3) based on [ 14 C]uridine influx experiments (18), and as illustrated in Fig. 8A for a single representative oocyte measured at a uridine concentration of 5 M, the Na ϩ activation curve was sigmoidal with a Hill coefficient (n) consistent with an apparent Na ϩ /nucleoside coupling stoichiometry of 2:1 (see also Fig. 6D). Both the apparent affinity for Na ϩ (K 50 ) and the maximal current (I max ) increased as the external concentration of uridine was raised (Table II). This pattern resembled that found for hCNT1 (32) and was consistent with a sequential mechanism of transport in which Na ϩ binds to the transporter first, increasing its affinity for nucleoside, which then binds second (37)(38)(39)(40). Parallel flux experiments with [ 14 C]uridine produced similar findings (data not shown).
The relationship between hCNT3 current evoked by 100 M uridine and external pH in the absence of Na ϩ (ChCl, pH 8.5-4.5) was also investigated (Table II). As illustrated for the representative oocyte in Fig. 8B, and in contrast to the sigmoidal activation curve observed for Na ϩ , a plot of current versus H ϩ concentration was hyperbolic with a Hill coefficient (n) consistent with a H ϩ /nucleoside coupling stoichiometry of 1:1. Parallel [ 14 C]uridine influx experiments revealed similar hyperbolic H ϩ activation kinetics (data not shown). Apparent K 50 FIG. 7. Kinetic properties of Na ؉ -and H ؉ -coupled hCNT3. Initial rates of [ 14 C]uridine uptake (20°C, 1-min flux) were measured in Na ϩ -containing (100 mM NaCl) and choline-containing (100 mM ChCl) transport media at either pH 7.5 (A and B, respectively) or pH 5.5 (C and D, respectively) in oocytes injected either with water alone (open circles) or with water containing hCNT3 RNA transcripts (solid circles). All of the fluxes were performed on the same batch of oocytes used on the same day. Kinetic parameters derived from these data for the hCNT3-mediated component of transport (uptake in RNA transcript-injected oocytes minus uptake in water-injected oocytes) are presented in Table I. values for H ϩ and Na ϩ differed by four orders of magnitude (480 nM and 2.4 -5.9 mM, respectively) (Table II).
The first series of experiments was performed in Na ϩ -containing transport medium at pH 8.5, and at holding potentials of Ϫ30, Ϫ50, and Ϫ90 mV to determine the Na ϩ /nucleoside coupling ratio and its voltage dependence. At a holding potential of Ϫ30 mV, the linear correlation between uridine-dependent charge and uridine accumulation gave a stoichiometry of 1.4 (Fig. 9A) (Table III). This increased to 1.6 at Ϫ50 mV and 1.9 at Ϫ90 mV (Fig. 9, B and C) (Table III). In marked contrast, parallel experiments performed in Na ϩ -free ChCl transport medium at pH 5.5 to determine the H ϩ /nucleoside coupling ratio at V h Ϫ30, Ϫ50, and Ϫ90 mV found voltage-independent stoichiometries in the range 0.92-1.1 (Fig. 9, D-F) (Table III). The same analysis was also performed in the presence of Na ϩ and H ϩ (NaCl, pH 5.5). As summarized in Table III, coupling ratios were voltage-independent and in the range 1.9 -2.0.
Charge-to-Na ϩ Stoichiometry-The relationship between uridine-evoked charge influx (picomoles) and 22 Na ϩ uptake (picomoles) was measured in five oocytes clamped at Ϫ90 mV (Fig. 10). A linear fit of the data gave a regression line with a slope of 0.97, indicating that one net inward positive charge was transported for every Na ϩ ion cotransported with uridine into the cell (Table III).
Characterization of hCNT3/hCNT1 Chimeras-hCNT3 and hCNT1 are 48% identical and 57% similar in predicted amino acid sequence, with the strongest residue conservation in TMs of the C-terminal halves of the proteins (Fig. 11A). The major differences lie in the putative N-and C-terminal tails and in the first three TMs. To localize domains involved in cation recognition, a chimera (hCNT3/1), in which the C-terminal half of hCNT3 (incorporating TMs 7-13) was replaced with that of hCNT1, was constructed. The splice site between the two proteins following hCNT3 residue Lys 314 was engineered at the beginning of the putative extramembraneous loop prior to TM 7 to divide the proteins into two approximately equal halves FIG. 8. Na ؉ and H ؉ activation of hCNT3. A, Na ϩ activation curve in a single representative hCNT3-producing oocyte measured in transport medium containing 0 -100 mM NaCl (pH 8.5) at a uridine concentration of 5 M and a membrane potential of Ϫ50 mV. Currents were normalized to the fitted I max (ϮS.E.) of 53 Ϯ 3 nA. The inset in A is a Hill plot of the data. The K 50 was 7.0 Ϯ 0.6 mM, and the Hill coefficient was 1.6 Ϯ 0.2. No currents were observed in control water-injected oocytes. Results from mean Na ϩ -activation experiments performed at different uridine concentrations are presented in Table II. B, H ϩ -activation curve in a single representative hCNT3-producing oocyte measured in Na ϩfree transport medium (100 mM ChCl) at pH 8.5-4.5 and a membrane potential of Ϫ50 mV. Currents were normalized to the fitted I max of 120 Ϯ 11 nA. The inset in B is a Hill plot for the data. The K 50 was 470 Ϯ 99 nM, and the Hill coefficient was 0.68 Ϯ 0.06. No currents were observed in control water-injected oocytes. Mean H ϩ -activation data from six individual oocytes are summarized in Table II.  Fig. 7 (A-D). b In transport media containing either 100 mM NaCl or 100 mM ChCl.

TABLE II Na ϩ and H ϩ activation kinetics of hCNT3
Hill coefficients (n) and apparent affinities (K 50 ) for Na ϩ were determined from Na ϩ concentration response curves (0 -100 mM NaCl, pH 8.5) in hCNT3-producing oocytes measured at uridine concentrations of 5, 25, and 100 M (see Fig. 8A for a representative experiment at 5 M uridine). Those for H ϩ were determined from H ϩ concentration response curves (pH 8.5-4.5) measured in Na ϩ -free transport medium (100 mM ChCl) at a uridine concentration of 100 M (see Fig. 8B for a representative experiment). Values were obtained from fits to data from individual oocytes normalized to the fitted I max value obtained for that cell and are presented as means Ϯ S.E. I max values (nA) (ϮS.E.) in the presence of Na ϩ were determined separately at a saturating concentration of uridine (100 M) with 100 mM external Na ϩ (pH 8.5) in oocytes from a single batch of cells used on the same day. The numbers in parentheses denote the number of oocytes. The membrane potential was Ϫ50 mV.  (6) a The apparent K 50 for Na ϩ is in millimolar, whereas that for H ϩ is in nanomolar. and to minimize disruption of native TMs and loops. The resulting hCNT3/1 chimeric protein was functional when produced in Xenopus oocytes and exhibited hCNT1-like substrate selectivity for influx of 20 M radiolabeled pyrimidine and purine nucleosides (NaCl, pH 7.5): uridine, thymidine, cytidine Ͼ Ͼ adenosine, and no detectable transport of guanosine or inosine (Fig. 11B). The reciprocal chimera (hCNT1/3), representing a 50:50 construct incorporating the N-terminal half of hCNT1 and the C-terminal half of hCNT3, was non-functional and not studied further. As shown in Fig. 11C, hCNT3/1 was Na ϩ -dependent, but H ϩ -independent, demonstrating that the structural features determining H ϩ coupling reside in the Cterminal half of the protein. Similar to hCNT1 (14, 32), the relationship between hCNT3/1-mediated [ 14 C]uridine influx and Na ϩ concentration at pH 7.5 was hyperbolic with a Hill coefficient (n) of 1.0 Ϯ 0.1 (Fig. 11D), a value consistent with a Na ϩ /nucleoside coupling stoichiometry of 1:1.

DISCUSSION
The CNT protein family in humans is represented by three members, hCNT1, hCNT2, and hCNT3, corresponding to concentrative nucleoside transport processes cit, cif, and cib, respectively. hCNT3 is a transcriptionally regulated electrogenic transport protein that, unlike hCNT1 and hCNT2, transports a broad range of pyrimidine and purine nucleosides and nucleoside drugs (18). hCNT3 and its mouse ortholog mCNT3 are FIG. 9. Stoichiometries of Na ؉ /uridine and H ؉ /uridine cotransport by recombinant hCNT3. Charge to [ 14 C]uridine uptake ratio plots were generated with 9 to 10 different hCNT3-producing oocytes in Na ϩ -containing transport media (100 mM NaCl, pH 8.5) at membrane holding potentials V h ϭ Ϫ30 mV (A), V h ϭ Ϫ50 mV (B), and V h ϭ Ϫ90 mV (C). Similar conditions were used for D-F (n ϭ 9) except that Na ϩ in the transport buffer was replaced by choline (100 mM ChCl), and the medium was acidified to pH 5.5. Integration of the uridine-evoked current was used to calculate the net cation influx (charge) and was correlated to the net [ 14 C]uridine influx (flux). Linear regression analysis of the data for each plot is indicated by the solid line. The dashed line indicates a theoretical 2:1 charge:uptake ratio in A-C (presence of Na ϩ ) and a 1:1 charge:uptake ratio in D-F (presence of H ϩ ). Linear fits passed through the origin. Stoichoimetries (ϮS.E.) obtained from these data and from corresponding experiments performed in Na ϩ -containing transport media (100 mM NaCl) at pH 5.5 (i.e. in the presence of both Na ϩ and H ϩ ) are summarized in Table III. more closely related in sequence to the prevertebrate hagfish transporter hfCNT (30) than to mammalian CNT1/2, and thus form a separate CNT subfamily. The present study utilized heterologous expression in Xenopus oocytes in combination with electrophysiological, radioisotope flux, and chimeric experiments to characterize the selectivity, mechanism, energetics, and structural basis of hCNT3 cation coupling. Parallel experiments with mCNT3 confirmed the general applicability of the reported findings.
Unlike hCNT1 and hCNT2, which are largely Na ϩ -specific, hCNT3 exhibited Na ϩ , H ϩ , and Li ϩ dependences, all three cations supporting electrogenic uridine influx. In this regard, hCNT3 resembles the mammalian concentrative glucose transporters SGLT1 and SGLT3, the bacterial MelB melibiose transporter, and the mammalian Na ϩ /dicarboxylate cotransporter SDCT1/NaDC-1, all of which can also utilize Na ϩ , H ϩ , or Li ϩ to drive cellular accumulation of substrate (42)(43)(44)(45)(46). Consistent with an intracellular oocyte pH of 7.3-7.6 (47), nucleosideevoked H ϩ currents were minimal at external pH values of 7.5 or higher. Use of CCCP to dissipate the imposed H ϩ electrochemical gradient across the cell membrane (48,49) decreased the uridine-evoked current in acidified Na ϩ -free medium by ϳ60%, confirming that hCNT3 is coupled to the proton-motive force. H ϩ and Li ϩ coupling within the CNT3/hfCNT subfamily is unique to h/mCNT3 and is not shared by hfCNT (30). In contrast, some other CNTs function exclusively as H ϩ /nucleoside symporters, including NupC from E. coli, CeCNT3 from C. elegans, and CaCNT from C. albicans (26 -28).
Na ϩ /nucleoside and H ϩ /nucleoside symport by hCNT3 exhibited markedly different selectivity characteristics for physiological nucleosides and therapeutic nucleoside drugs, suggesting that Na ϩ and H ϩ binding induce cation-specific conformational changes in the hCNT3 substrate-binding pocket and/or translocation channel. For H ϩ -coupled hCNT3, this was reflected in markedly reduced transport activity for thymidine, cytidine, adenosine, and inosine, and inability to transport guanosine, AZT, and ddC. In a possibly related phenomenon, functional studies with microglia have shown that an inwardly directed H ϩ gradient can inhibit AZT uptake (50). Microglia have cib-type activity as a major component of their nucleoside transport machinery (51).
Consistent with results of recent studies of hCNT1 (32), the proton/myo-inositol cotransporter (38), and SGLT1 (39), hCNT3 kinetic experiments revealed an ordered binding mechanism. Na ϩ removal increased the transporter's K m for uridine by more than 30-fold, this being accompanied by a smaller (1.6-fold) increase in V max . Limiting the concentration of Na ϩ can therefore be overcome by increasing the concentration of uridine to reach similar maximal rates of transport (37)(38)(39)(40). In contrast, limiting the concentration of uridine reduced both the apparent affinity of hCNT3 for Na ϩ and the maximal current. Na ϩ therefore binds to hCNT3 first, followed by nucleoside. Like SGLT1 (44), the apparent affinity of hCNT3 for H ϩ was four orders of magnitude higher than for Na ϩ . H ϩ and Na ϩ binding to SGLT1 also lead to cation-specific conformational changes, which, like hCNT3, were reflected in a decrease in sugar-binding affinity and transport efficiency of the H ϩ -coupled transporter (45). In the case of hCNT3, substitution of H ϩ for Na ϩ resulted in a 6-fold change in uridine apparent affinity, and a decrease of ϳ70% in V max :K m ratio (Table I). Unlike hCNT3, however, H ϩ -coupled SGLT1 exhibited only modest changes in sugar specificity compared with the Na ϩ -coupled transporter (45). Similar to SGLT1, hCNT3-mediated Na ϩ / nucleoside and H ϩ /nucleoside symport were voltage-dependent (35,36). The apparent affinity of hCNT3 for uridine was voltage-insensitive at high external Na ϩ concentrations, but voltagedependent when the concentration of Na ϩ was reduced, suggesting that the voltage dependence of the transporter's apparent affinity for uridine may be due to the voltage dependence of Na ϩ binding (35,36). This was supported by experiments showing a similar voltage dependence of the apparent affinity of hCNT3 for Na ϩ . As in the case of SGLT1, these results are indicative of the presence of an ion well effect (35,36), with pre-steady-state electrophysiological studies suggesting that Na ϩ binds to hCNT3 ϳ 40% within the membrane electric field (52). Na ϩ /nucleoside and H ϩ /nucleoside I-V curves and the effect of membrane potential on I max values suggest that membrane potential also influences carrier translocation. This is consistent with the fact that the driving force for an electrogenic transporter is dependent on not only the gradients of substrate and cotransported ion, but also on the membrane potential.
Based on indirect Hill-type analyses of the relationship between nucleoside influx and Na ϩ concentration, Na ϩ /nucleo- FIG. 10. Charge-to-Na ؉ stoichiometry of hCNT3. The charge to 22 Na ϩ uptake plot was generated from five different hCNT3-producing oocytes at a membrane potential of Ϫ90 mV. The sodium and uridine concentrations were 1 mM (pH 8.5) and 200 M, respectively. Linear regression analysis of the data is indicated by the solid line (see Table  III for the fitted slope). The dashed line indicates a theoretical charge: uptake ratio of 1:1. The linear fit passed through the origin.  Fig. 9. b From Fig. 10. c The numbers in parentheses denote the number of oocytes.
FIG. 11. Transport properties of chimera hCNT3/1. A, topographical model of hCNT3 and hCNT1. Potential membrane-spanning ␣-helices are numbered, and putative glycosylation sites in predicted extracellular domains in hCNT3 and hCNT1 are indicated by solid and open stars, respectively. Residues identical in the two proteins are shown as solid circles. Residues corresponding to insertions in the sequence of hCNT3 or hCNT1 are indicated by circles containing "ϩ" and "Ϫ" signs, respectively. The arrow represents the splice site used for construction of the chimera. B, uptake of radiolabeled nucleosides by chimera hCNT3/1. Nucleoside influx (20 M, 20°C, 1-min flux) was measured in transport medium containing 100 mM NaCl at pH 7.5. Mediated transport was calculated as uptake in RNA transcript-injected oocytes minus uptake in control water-injected oocytes. C, radiolabeled uridine influx (20 M, 20°C, and 1-min flux) by hCNT1, hCNT3, and hCNT3/1 was measured in transport medium containing 100 mM NaCl at pH 7.5 (black bars) or in Na ϩ -free transport medium (100 mM ChCl) at both pH 5.5 (gray bars) and 7.5 (open bars). Mediated transport was calculated as uptake in RNA-injected oocytes minus uptake in control waterinjected oocytes. D, influx of [ 14 C]uridine (20 M, 20°C, and 1-min flux) measured as a function of Na ϩ concentration at pH 7.5 using choline as Na ϩ substitute in oocytes injected with water alone (open circles) or with water containing hCNT3 RNA transcripts (solid circles). The inset in B is a Hill plot of the mediated data (Hill coefficient (n) and Na ϩ apparent affinity (K 50 ) presented in the text). Values in B-D are means Ϯ S.E. of 10 -12 oocytes. Each experiment in B-D was performed on cells from single batches of oocytes used on the same day. side stoichiometries of 1:1 have been described for cit and cif transport activities in different mammalian cells and tissues (7-9) and for recombinant rCNT1 and hCNT1 produced in Xenopus oocytes (32,53). Although Larrá yoz et al. (31) have reported a hCNT1 Na ϩ /nucleoside coupling ratio of 2:1 based on results of direct charge versus [ 3 H]uridine and 22 Na ϩ uptake studies, we have recently confirmed that the stoichoimetry of hCNT1 is 1:1 (32). H ϩ -dependent CaCNT also exhibits a coupling ratio of 1:1 (28). In marked contrast, there is evidence from Na ϩ -activation studies of mammalian cib transporters (14,18,54) and from charge versus uridine uptake experiments with hagfish hfCNT (30) that members of the CNT3/hfCNT subfamily have a coupling ratio of 2:1. The charge versus uridine and charge versus Na ϩ uptake experiments reported here for hCNT3 confirmed this stoichiometry. The Hill coefficient for Na ϩ activation of hCNT3 is close to 2, implying strong cooperativity between the two Na ϩ -binding sites (55,56).
Similar to SGLT1 (39,57,59), but unique among the other CNTs that have been examined to date (hCNT1, hfCNT, and CaCNT) (28,30,32), the experimentally determined Na ϩ /nucleoside coupling ratio of hCNT3 was voltage-dependent, increasing progressively to its maximum value of 2:1 as the membrane potential became more negative. As in the case of SGLT1 (59), this may reflect an effect of membrane potential on Na ϩ dissociation from the cytoplasmic face of the transporter, with a consequent reduction in Na ϩ recycling back to the external surface of the membrane. Other transporter families also have individual members with different cation coupling ratios. For example, members of the SGLT family have been described with 1:1 and 2:1 Na ϩ /glucose coupling ratios (1:1 for SGLT2 and 2:1 for SGLT1/3) (39,57,59,60). Similarly, the PepT1 and PepT2 proton-linked peptide transporters have 1:1 and 2:1 H ϩ /peptide coupling ratios, respectively (61). Mechanistically, the cation-to-nucleoside coupling ratio determines the energetic cost of transport and sets the thermodynamic limit to the transmembrane nucleoside gradients that can be achieved. The concentrative capacity of hCNT3 is therefore greater than that of either hCNT1 or hCNT2.
In marked contrast to SGLT1, where both Na ϩ and H ϩ have the same 2:1 cation/sugar stoichiometries (44,59,62), charge/ uptake analyses revealed an hCNT3 Na ϩ /nucleoside coupling ratio of 2:1 versus 1:1 for H ϩ . Unlike Na ϩ , the H ϩ coupling ratio was membrane potential-independent. Charge/uptake experiments with both Na ϩ and H ϩ present together (Na ϩ -containing transport medium at pH 5.5) revealed features intermediate between Na ϩ or H ϩ alone (2:1 coupling ratio and voltageindependent), suggesting that both cations contributed to the driving force under these conditions. The 2:1 charge/uptake coupling ratio under these conditions implies that one of the two Na ϩ binding sites is shared by H ϩ . Because proton-activation experiments gave a Hill coefficient consistent with a 1:1 H ϩ /nucleoside coupling ratio, it is unlikely that there exists a second (recycled) H ϩ bound to hCNT3. The hCNT3 cation binding site shared by Na ϩ and H ϩ is likely to be the same as that responsible for single-site cation coupling in CNTs that are either exclusively H ϩ -dependent (CaCNT, NupC) (26,28) or Na ϩ -dependent (hCNT1/2) (7-9, 32, 53).
We interpret our results for cation coupling of hCNT3 in terms of the conformational equilibrium model of secondary active transport developed by Krupka (63,64). This modified ordered binding model of secondary active transport alleviates the stringent sequential carrier states of earlier models and instead allows for flexible cation interactions such as those observed for Na ϩ and H ϩ coupling of hCNT3. Because the transporter can accept two different solutes, cation (A) and nucleoside (S), it is proposed to exist in two inwardly facing or outwardly facing conformational states: one that binds cation only (T i Ј and T o Ј) and one that binds both cation and nucleoside (T i Љ and T o Љ). Normally, the equilibrium between the two outwardly facing carrier states overwhelmingly favors the T o Ј form and requires the addition of cation (two Na ϩ ions or one H ϩ in the case of hCNT3) to "unlock" or open the nucleoside binding site (T o ЈA 7 T o ЉA), thereby promoting active transport. Both T o ЉS and T o ЉAS are considered mobile. The finding that binding of an H ϩ to one of the two Na ϩ -binding sites is sufficient to activate nucleoside transport presents an experimental paradigm to enable mutageneic dissection of amino acid residues contributing to each of the sites. Our 50:50 hCNT3/1 chimera demonstrated that the structural determinants of cation/nucleoside stoichiometry and H ϩ dependence reside in the C-terminal half of the protein. Hill-type analysis of Na ϩ /nucleoside coupling in a corresponding 50:50 chimera between hf-CNT and hCNT1 yielded similar results (30). The present finding that determinants of hCNT1 versus hCNT3 nucleoside selectivity also reside in the C-terminal half of the protein is consistent with previous mutagenesis experiments that identified residues in TMs 7 and 8 of hCNT1 that, when sequentially mutated to the corresponding residues in hCNT2, progressively changed the selectivity of the transporter from cit to cib to cif (29).
In conclusion, hCNT3 exhibited unique cation interactions with Na ϩ , H ϩ , and Li ϩ that are not shared by other members of the CNT protein family. Both indirect and direct methods indicated 2:1 and 1:1 cation/nucleoside stoichiometries for Na ϩ and H ϩ , respectively, and Na ϩ -and H ϩ -coupled hCNT3 exhibited markedly different selectivities for nucleoside and nucleoside drug transport. Location of hCNT3-specific cation interactions to the C-terminal half of the protein sets the stage for site-directed mutagenesis experiments to identify the residues involved. The ability of hCNT3 to couple nucleoside and nucleoside drug accumulation to H ϩ as well as Na ϩ cotransport may have physiological and pharmacological relevance in the duodenum and proximal jejunum where the pH of luminal contents can be relatively acidic. As well, there is a reported acidic microenvironment present on the surface of the intestinal epithelium (58).